GPT-5.6 Sol vs Kimi K2.5
Wins 8 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
GPT-5.6 Sol is the stronger all-rounder.Kimi K2.5 is cheaper.
Scores updated · 47 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- AgentsCarrying out multi-step tasks on its own60GPT-5.6 Sol6 of 6 tests
- CodingWriting and fixing software50GPT-5.6 Sol5 of 5 tests
- ReasoningHard problems that need careful thinking40GPT-5.6 Sol4 of 4 tests
- Images and chartsUnderstanding pictures, charts and video30GPT-5.6 Sol3 of 3 tests
- FactsGetting facts right instead of making them up31GPT-5.6 Sol3 of 4 tests
- MathCompetition and research-level math10GPT-5.6 Sol1 of 1 test
- Long documentsFinding answers in very long texts10GPT-5.6 Sol1 of 1 test
- Following instructionsDoing exactly what it is asked10GPT-5.6 Sol1 of 1 test
Math, long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
Kimi K2.5 costs 85% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where GPT-5.6 Sol pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+80.7points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+42.3points ahead
- Real work tasks from 44 professionsGDPVal+38.4points ahead
Where Kimi K2.5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+26.5points ahead
Every test, side by side
All 47 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.6 Sol
- Terminal-Bench 2.1GPT-5.6 Sol by 42.38845.7+42.3
- Terminal-Bench HardGPT-5.6 Sol by 31.165.934.8+31.1
- LMArena · WebDevGPT-5.6 Sol by 182 rating points16181436+182 rating
- SWE-bench ProGPT-5.6 Sol by 13.964.650.7+13.9
- SciCodeGPT-5.6 Sol by 8.157.149+8.1
AgentsGPT-5.6 Sol
- GDPValGPT-5.6 Sol by 38.455.617.2+38.4
- τ-Bench V3 · BankingGPT-5.6 Sol by 30.144.314.2+30.1
- AA ApexAgentsGPT-5.6 Sol by 28.439.911.5+28.4
- MCP AtlasGPT-5.6 Sol by 19.883.663.8+19.8
- OSWorld-VerifiedGPT-5.6 Sol by 19.78363.3+19.7
- BrowseCompGPT-5.6 Sol by 15.590.474.9+15.5
ReasoningGPT-5.6 Sol
- ARC-AGI-2GPT-5.6 Sol by 80.792.511.8+80.7
- CritPtGPT-5.6 Sol by 29.232.33.1+29.2
- Humanity's Last ExamGPT-5.6 Sol by 18.849.530.7+18.8
- GPQA DiamondGPT-5.6 Sol by 6.294.187.9+6.2
FactsGPT-5.6 Sol
- SimpleQA VerifiedGPT-5.6 Sol by 35.469.734.3+35.4
- AA-Omniscience · Non-hallucinationKimi K2.5 by 26.57.834.3+26.5
- AA-Omniscience · AccuracyGPT-5.6 Sol by 24.259.435.2+24.2
- Vectara HHEM hallucination ratelower is betterGPT-5.6 Sol by 1.812.414.2+1.8
Images and chartsGPT-5.6 Sol
- CharXiv (RQ)GPT-5.6 Sol by 8.385.877.5+8.3
- MMMU-ProGPT-5.6 Sol by 883.475.4+8
- LMArena · VisionGPT-5.6 Sol by 10 rating points12791269+10 rating
Following instructionsGPT-5.6 Sol
- IFBenchGPT-5.6 Sol by 2.572.770.2+2.5
Other results22 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- GDPval-AA v2GPT-5.6 Sol by 702 rating points17111009+702 rating
- CyberGymGPT-5.6 Sol by 43.284.541.3+43.2
- ARC-AGI-1GPT-5.6 Sol by 31.296.565.3+31.2
- Artificial Analysis Coding IndexGPT-5.6 Sol by 30.677.446.8+30.6
- ToolathlonGPT-5.6 Sol by 30.25827.8+30.2
- AA-OmniscienceGPT-5.6 Sol by 29.322-7.3+29.3
- AA Agentic IndexGPT-5.6 Sol by 28.850.521.7+28.8
- PaperBenchGPT-5.6 Sol by 2790.563.5+27
- AA IntelligenceGPT-5.6 Sol by 23.54723.5+23.5
- τ³-Bench BankingGPT-5.6 Sol by 18.83314.2+18.8
- BrowseComp (w/ Ctx)GPT-5.6 Sol by 14.589.474.9+14.5
- HLE (with tools)GPT-5.6 Sol by 14.364.550.2+14.3
- ResearchRubricsGPT-5.6 Sol by 14.373.859.5+14.3
- SWEBench Pro PublicGPT-5.6 Sol by 13.964.650.7+13.9
- τ²-Bench Telecom (AA run)Kimi K2.5 by 10.885.195.9+10.8
- charxiv_rq_with_pythonGPT-5.6 Sol by 9.187.878.7+9.1
- Global-MMLU-LiteGPT-5.6 Sol by 7.891.884+7.8
- vectara_answer_rateGPT-5.6 Sol by 6.999.192.2+6.9
- LVBenchGPT-5.6 Sol by 6.282.175.9+6.2
- vectara_avg_summary_lengthGPT-5.6 Sol by 19.1 rating points131.1112+19.1 rating
- vectara_factual_consistencyGPT-5.6 Sol by 1.887.685.8+1.8
- StrongREJECTKimi K2.5 by 198.599.5+1
Questions people ask
Which is better, GPT-5.6 Sol or Kimi K2.5?
GPT-5.6 Sol wins all eight areas we test: coding, agents, reasoning, facts, images and charts, math, long documents and following instructions. Kimi K2.5 wins none, but costs 85% less.
Which is better for coding?
GPT-5.6 Sol. It wins 5 of the 5 coding tests both models report; Kimi K2.5 wins none.
Which is cheaper?
GPT-5.6 Sol costs $4.00 per million input tokens and $20.00 per million output tokens; Kimi K2.5 costs $0.60 and $3.00. That makes Kimi K2.5 about 85% cheaper for the same work.
How do you compare the two?
We use the 47 benchmark tests both models have published scores on. The verdict counts the 25 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 22 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.